Main
Saturday
Nov142015

Building on a House of Cards: On-Demand Culture and Big Data

Chuck Tryon

 

[ PDF Version ]

 

One of the perverse pleasures of watching the Netflix original series House of Cards (2013–) is seeing the show’s antihero, Frank Underwood (Kevin Spacey), as he manipulates and murders his way from the position of House Majority Whip to the Presidency of the United States. Underwood seduces us into his worldview through monologues in which he directly addresses viewers, treating them as accomplices in his crimes. His sinister ability to manipulate the levers of power in Washington, D.C., might confirm what some of us already ostensibly know about politics: the game is rigged to benefit the powerful and opportunistic, especially those who are willing to discard their principles for a little more power. This political cynicism might have made the show too edgy for network television. However, House of Cards has been the centerpiece of Netflix’s foray into original programming, at least in terms of the subscription video-on-demand (SVOD) company’s self-promotion. As Netflix itself openly acknowledged, the company drew extensively from users’ data profiles to produce and promote the series to a wide variety of audiences. Netflix’s emphasis on House of Cards, I argue, can tell us quite a bit about the formation of an on-demand spectator, one that is constructed not just through interfaces but also through a wide range of industrial discourses, including Netflix’s own promotional discourses. In particular, this concept of on-demand spectatorship has been defined via Netflix’s careful engagement with and representation of the phenomenon of “Big Data” to analyze its subscribers’ viewing habits and taste preferences. But rather than focusing on how Netflix actually uses viewer data, I am more interested in seeing how their promotional discourses participate in the ongoing ideological construct of a highly individualized on-demand spectator—one that Netflix portrays as having access to unlimited consumer choice.

This paper grows out of an attempt at developing a more detailed understanding of how digital delivery has contributed to a reconsideration of the role of the movie and television audience. By thinking about how digital delivery platforms work and how they have been theorized through promotional articles and other forms of publicity, I argue that we might be able to reopen debates about media spectatorship and the role of industry discourse in shaping it. In general, the promise of unlimited consumer choice and flexibility offered by digital delivery has been the subject of almost relentless promotion and hype.[1] This reframing of televisual spectatorship is part of what Amanda D. Lotz has called the “post-network era,” in which audiences have access to an apparently limitless array of television channels and viewing options.[2] Alongside this emphasis on personalized choice is the promise that VOD users can receive recommendations tailored directly to their interests. In fact, Netflix famously sponsored a contest in 2009 inviting teams of researchers to compete to improve their recommendation algorithm by at least ten percent, with the winning group receiving a $1 million prize.[3] In essence, through its use of algorithmic recommendations, Netflix seems to promise us that it knows what we want to watch even before we do. But such promises have also been accompanied by wider concerns about issues ranging from concerns about protecting user privacy on the one hand to artificial fears about algorithms taking control of the creative process on the other.

This perception of unlimited choice can be fruitfully compared to ongoing gags about the lack of consumer options during the video store era. Recall, for example, George Costanza’s desperate attempts to secure the last copy of Breakfast at Tiffany’s at a local video store on an episode of Seinfeld or the satirical Onion video depicting people going on historical tours of Blockbuster Video stores, where they encounter the “hardships” of renting videos in an age before VOD.[4] In both cases, physical copies of movies or TV series lead to frustration and inconvenience, as our desires to watch what we want on our own schedules are thwarted. In other Seinfeld episodes, Elaine relies on the guidance of a precocious video store clerk who advises her on what movies to see, a plot that helps to illustrate the choice paralysis many video store consumers felt when they went to rent a movie for a night’s entertainment.[5] Digital delivery has been promoted, in part, as offering a solution to these types of challenges.

The shift to digital delivery has prompted a wide range of efforts to make sense of how the on-demand viewer (or what Netflix Chief Content Officer Ted Sarandos has called the “addressable audience”) will navigate these choices.[6] The typical image, at least within the advertising of companies like Comcast or Time Warner, has been one of unlimited choice producing either individual empowerment or family harmony, as individual viewers are able to control what they watch and how they watch it. In fact, in some advertisements, spouses are shown together in bed, but watching separately, often on personalized media devices like tablets or smart phones, able to avoid conflict through personalized screen experiences. However, alongside these models of on-demand spectatorship, we also find a somewhat more unsettling image, one in which media consumers are depicted as puppets of the very tools that provide users with the illusion of unlimited choice. In this scenario, consumers are merely data points within a larger network of media producers, distributors, and exhibitors that use complex computer algorithms to analyze rental or purchasing habits in order to tailor content directly to specific consumers in a way that (according to some observers) undermines consumer choice. This debate over the implications of Big Data serves as a reminder that we need to explore how the concept of the algorithm has been used to make sense of both the emerging on-demand viewer and the future of media distribution itself.

The concept of what some critics refer to as the monitorable spectator has been discussed in media scholarship. As Mark Andrejevic observes, “customization . . . increases the demand for demographic information to a new level; it creates new markets for the fruits of increasingly extensive and intensive forms of consumer surveillance.”[7] In fact, in Andrejevic’s estimation, invitations to “participate” with media texts are a form of control, and he flatly suggests that interactivity—whether it entails live-tweeting a favorite show or offering ratings of movies on Netflix or Hulu—is, in essence, “a means of exploiting audience labor.”[8] Even practices such as telling Netflix who is using the account at any given time, much less linking your Netflix and Facebook accounts, can provide data that could be used to monitor how viewers are using the service. Yet in some ways, Netflix promotes the perception that there is an implicit benefit to sacrificing privacy. When searching for personalized recommendations for myself, I must often wade through past selections made by my teen stepchildren and other selections used to serve as an electronic babysitter for our toddler, making it tempting to answer Netflix’s request to know who is watching at any given moment. To make sense of how these new forms of monitoring operate, José Van Dijck argues that an individual platform—such as Netflix’s website—should be understood as increasingly structuring our social interactions, including how we seek out modes of entertainment in ways that explicitly “reflect the platform owner’s strategic choices.”[9] In both cases, these new spectatorship practices are identified not with unlimited choice but with constraints, limitations, and in some cases, concern that users will be manipulated by the platforms they use and the underlying algorithms and interfaces that structure their use. As Van Dijck describes it, “algorithms increasingly determine what we like, want, know, or find.” [10] I’m certainly sympathetic to many of these arguments. For many viewers—especially in the post-video store era—if something is not on Netflix, it might as well not even exist. And even if it is on Netflix, previous viewing choices may make it harder for a user to discover that text. This practice of generating detailed viewer profiles was partially behind Netflix’s “Flixie” awards, in which it gave awards to movies and TV shows in its catalog based on humorously defined categories of viewing practices, such as best hangover movie, tantrum tamer, and bromance. Implicit in each of these categories are characteristics such as age, gender, and even the time of viewing (hangover movies would likely be watched on weekend afternoons). In some cases, Netflix’s use of data has inspired a backlash—as when personal information about individual users mistakenly went public—but in other cases, it has instead served as a means of promoting a personalized entertainment experience.

It was precisely this ambivalence that animated Andrew Leonard’s Salon article on Netflix’s remake of the British political drama House of Cards (BBC, 1990), a text that opened up an examination of the mainstream accounts of the concept of Big Data.[11] Leonard’s article proposes that Netflix is able to log literally hundreds of millions of “events,”—decisions to pause, fast-forward, or stop watching a video—on a daily basis, providing them with a massive pool of data that could be used not simply to determine which shows are most popular or which shows a particular viewer might like, but also to shape creative decisions about content. Leonard, responding to the popularity and critical acclaim of the Netflix original series House of Cards, offers an apparently sinister portrait of Netflix’s use of data analysis in order to assess how, when, and where people are watching. Thus, for Leonard, spectatorship becomes a monitorable activity, one in which viewers are tracked and then analyzed via Netflix’s algorithms in order to detect patterns that might be of use. In fact, Leonard points to the decision to obtain the rights to the House of Cards remake as being the direct result of the company’s use data analysis. Yet at the same time, Leonard’s account also helps to reinforce its reputation as being able to provide audiences what they want from a streaming video service. Citing data provided by Netflix, Leonard reports that the company detected a strong correlation between people who enjoyed the original British series and fans of the actor Kevin Spacey and the director David Fincher, and goes on to suggest that the decision to invest $100 million in the series was shaped by the company’s awareness of this potential demand. Thus, Leonard’s depiction almost seemed to merge platform with show, turning Big Data into a sinister puppet-master along the lines of Kevin Spacey’s ambitious political leader. Such concerns about Netflix’s use of data become a little more complicated, however, when we recall, as Timothy Havens has pointed out, that Netflix did not actually produce the first season of the show but instead purchased rights to an existing property.[12]

However, the dystopian vision of the dominance of data continues to retain enormous power. For example, Derrick Harris of GigaOm spelled out just how much data Netflix is collecting and which kinds of information might shape how the company addresses its users through its interfaces and recommendation algorithms. In examining its millions of users, Netflix was able to sift through 30 million plus “plays” per day, as well as all uses of pause, rewind, and fast-forwarding features; 4 million user ratings per day; 3 million searches per day; and information on the geographic location of the user and the time and day of the week in which specific content was viewed. As Harris observes, Netflix was able to learn (among many other things) that people are more likely to watch TV shows during the week and a larger number of movies on weekends. In addition, they picked up on the patterns of binge viewing that helped give rise to their decision to release all thirteen episodes of the first season of House of Cards simultaneously, a strategy that was replicated with Arrested Development, Orange is the New Black, and most other Netflix original series.[13] This use of data also allowed Netflix to craft trailers specifically for the different audiences that it imagined it was addressing. As David Carr reports, Netflix cut several different trailers to promote House of Cards: Kevin Spacey fans were shown a trailer that depicted his performance, while fans watching what Netflix defined as female-oriented content would see clips depicting interactions between female characters, and art cinema fans were addressed with a trailer that showcased Fincher’s command as a director, even though Fincher only directed two of the episodes from the first season. [14] Of course, the practice of creating trailers for specific audiences is nothing new. In fact, there is a long history of creating trailers or advertisements for specific anticipated audiences based on demographics, geographical location, or other traits. Jonathan Gray, in particular, has been attentive to the ways in which promotional paratexts might shape our expectations of what a TV show or movie is about.[15] However, Netflix promoted their use of multiple trailers as a way of highlighting the idea that they were providing viewers with personalized viewing experiences, a practice that aligned them with wider discourses of personalization associated with digital delivery.[16]

Beyond the substantial privacy concerns, for Leonard, the merger of Big Data and VOD threatens to undermine the creative process, potentially leading to what might be called a dictatorship of the data, in which recommendation algorithms do all the work of casting movies and creating storylines designed to have mass (or at least broad niche) appeal. And more chillingly, Leonard argues that companies that use this data will have the ability to “know more about us than we know ourselves, and will be able to craft techniques that push us toward where they want us to go, rather than where we would go by ourselves if left to our own devices.” Leonard’s remarks certainly overstate the degree to which data may control future storytelling decisions. As Havens notes, Netflix acquired the rights to the American remake of House of Cards, making it less a product of their algorithm than something that was purchased “based upon the interpretations of Netflix programmers.”[17] Thus, rather than a significant change in TV programming, the era of digital delivery offers a degree of continuity with the past as programmers continue to rely on whatever data is available to them to make choices about what to program and how to promote it. Programmers still make choices about what shows to purchase based on their interpretations of data—whether that involves test screenings or statistical reports about how people watch.

By way of a conclusion, I’d like to complicate some of these arguments about the monitorable viewer. While algorithms have intensified the ability to analyze viewer activity, Big Data is far less effective at analyzing why people are watching. In fact, as Viktor Mayer-Schönberger and Kenneth Cukier, the authors of Big Data, admit, “knowing why might be pleasant, but it’s unimportant for stimulating sales.”[18] In this sense, algorithms cannot account for precisely how users are making sense of the content they are watching, and for this reason, Netflix’s predictions may tell us little about the constellations of movies and TV shows that inhabit our viewing histories. Choosing to watch something does not indicate an endorsement of that text, a detail that may challenge even the most complex algorithm. Further, there is quite a bit of popular skepticism about the success of Netflix’s ability to make pertinent recommendations. This sentiment found expression in a widely circulated Onion video in which Netflix “announces” that they will be issuing a “browse endlessly” plan for $5.99 a month, allowing people to browse titles without ever choosing one.[19] Thus, while it might be easy to view Netflix as a puppet master in the mold of Frank Underwood, it’s also important to recognize the role of promotional discourse and industry analysis in shaping perceptions about how digital delivery is altering the practices of the on-demand spectator.

 

Notes

[1] Chuck Tryon, “‘Make any Room Your TV Room:’ Digital Delivery and Media Mobility,” Screen 53, no. 3 (2012): 287–300.

[2] Amanda D. Lotz, The Television Will Be Revolutionized (New York: New York University Press, 2007).

[3] Eliot Van Buskirk, “How the Netflix Prize Was Won,” Wired, 22 September 2009, www.wired.com/2009/09/how-the-netflix-prize-was-won/. Notably, despite the enormous resources and publicity given to the Netflix contest, very little of the winning algorithm was ever implemented, in part due to the company’s shift from a DVD-by-mail service into a streaming platform. See Casey Johnston, “Netflix Never Used Its $1 Million Algorithm Due to Engineering Costs,” Ars Technica, 13 April 2012, arstechnica.com/gadgets/2012/04/13/netflix-never-used-its-1-million-algorithm-due-to-engineering-costs/.

[4] “Historic Blockbuster Store Offers Glimpse of How Movies Were Rented in the Past,” The Onion, 12 May 2008, www.theonion.com/video/historic-blockbuster-store-offers-glimpse-of-how-m,14233/.

[5] For a discussion of representations of video store culture, see Daniel Herbert, Videoland: Movie Culture at the American Video Store (Berkeley: University of California Press), 2014.

[6] David Lieberman, “Netflix Unfazed by Growing Competition by Amazon, Execs Say,” Deadline Hollywood, 13 September 2012, deadline.com/2012/09/netflix-competition-amazon-ted-sarandos-335796/.

[7] Mark Andrejevic, iSpy: Surveillance and Power in the Interactive Era (Lawrence: University of Kansas Press, 2007), 13.

[8] Andrejevic, iSpy, 32.

[9] José van Dijck, The Culture of Connectivity: A Critical History of Social Media (Oxford: Oxford University Press, 2013), 29.

[10] van Dijck, The Culture of Connectivity, 37. Emphasis in original.

[11] Andrew Leonard, “How Netflix is Turning Viewers into Puppets,” Salon, 1 February 2013, www.salon.com/2013/02/01/how_netflix_is_turning_viewers_into_puppets/.

[12] Timothy Havens, “Media Programming in an Era of Big Data,” Media Industries 1, no. 2 (2014).

[13] Derrick Harris, “Netflix Analyzes a Lot of Data about Your Viewing Habits,” GigaOM, 14 June 2012, gigaom.com/2012/06/14/netflix-analyzes-a-lot-of-data-about-your-viewing-habits/.

[14] David Carr, “Giving Viewers What They Want,” Decoder Blog, New York Times, 24 February 2013, www.nytimes.com/2013/02/25/business/media/for-house-of-cards-using-big-data-to-guarantee-its-popularity.html.

[15] Jonathan Gray, Show Sold Separately: Promos, Spoilers, and Other Media Paratexts (New York: New York University Press), 2010.

[16] Tryon, “‘Make any Room Your TV Room.’”

[17] Havens, “Media Programming.”

[18] Viktor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution That Will Transform How We Live, Work, and Think (New York: Houghton Mifflin, 2013), 52.

[19] “Netflix Introduces New Browse Endlessly Plan,” The Onion, 18 February 2014, www.theonion.com/video/netflix-introduces-new-browse-endlessly-plan,35308/.

 

Chuck Tryon is an Associate Professor of English at Fayetteville State University. He has published two books, Reinventing Cinema: Movies in the Age of Media Convergence and On-Demand Culture: Digital Delivery and the Future of Movies, both with Rutgers University Press.

Reader Comments

There are no comments for this journal entry. To create a new comment, use the form below.

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>